Inconsistent behavior of FixedDelayRetry in combination with queue messages using C#

[plamber]

Check for a solution in the Azure portal

Hello, I am experiencing some unexpected behavior using following attribute on a Queue trigger written on C#.

[FixedDelayRetry(2, "00:01:00")]

My expectation is that the function does following:

• Function fails
• Waiting time for 1 minute
• Function fails a second time -> Goes to the next stage

Under next stage I understand that in case of a queue trigger it follows the maxDequeueCount value and follows the process above following the DequeueCount value.

I tried to configure host.json as follows:

{
  "version": "2.0",
  "queues": {
    "maxDequeueCount": "1",
    "maxPollingInterval": "00:00:02"
  },
  "logging": {
    "applicationInsights": {
      "httpAutoCollectionOptions": {
        "enableW3CDistributedTracing": true
      },
      "samplingSettings": {
        "isEnabled": true,
        "excludedTypes": "Request"
      }
    }
  }
}

With this configuration I am expecting that it tries to follow only the FixedDelayRetry and puts after 2 failures the queue message into poison queue.

It indeed follows the FixedDelayRetry twice with the specified amount of time. But on the second failure it immediately triggers the queue 2 additional times without waiting:

• First run - First attempt (failure)
• Waits a minute
• First run - Second attempt (failure)
• Second run - First attempt (failure)
• Second run - Second attempt (failure)
• Goes to poison

But based on my configuration my expectation is that it just executes the first run and then putting the message immediately in the poison queue. Even if the second run would be ok, it is no more considering the configured fixeddelayretry timespan. If I I configure a maxDequeueCount of 2 I would expect following behavior for example:

• First run - First attempt (failure)
• Waits a minute
• First run - Second attempt (failure)
• Waits a minute
• Second run - First attempt (failure)
• Waits a minute
• Second run - Second attempt (failure)
• Goes to poison

Please provide the following:

• Function App version: Local development
• Programming language used: C#
• Microsoft.Azure.Functions.Extensions: v.1.1.0
• Microsoft.NET.Sdk.Functions: 3.0.11
• Please note that I am using dependency injection for the functions. I have a simple example of the queue below

Source

```csharp public class WebhookQueue { //constructor with dependency injection [FunctionName(nameof(WebhookQueue))] [FixedDelayRetry(2, "00:01:00")] public async Task RunAsync([QueueTrigger(IEasyLifeQueueEntry.WebhookEvent, Connection = "AzureWebJobsStorage")] string input, ILogger log) { throw new Exception() } } ```

Thank you for your feedback, Patrick

[v-anvari]

Hi @plamber, This feature is in preview mode and the team is actively working towards improving documentation on the expected behavior for functions that use retry and deploy to Azure. You can find related information here - https://github.com/Azure/azure-webjobs-sdk/issues/2624

[plamber]

Hi @v-anvari , Thank you for your feedback. I checked your reference but I am not understanding if what I described is as expected or not.

Could you share more insights?

Thank you

[v-anvari]

Hi @plamber, There are known in consistent behavior with the Retry policy while using C#. The team is working on the Product Fix. You can find the Document communication that had happened earlier here- https://github.com/MicrosoftDocs/azure-docs/issues/71628

[hezite]

are there any updates on the issue?